Using the AR-V7 biomarker to determine treatment in metastatic castrate resistant prostate cancer, a feasibility randomised control trial, conclusions from the VARIANT trial

Introduction

Prostate cancer is the most diagnosed malignancy in the United Kingdom (UK) and the second most common cause of cancer mortality¹. Whilst overall survival rates are high, metastatic prostate cancer is incurable with poor five-year survival rates². Treatment for metastatic prostate cancer includes medical or surgical castration, the former consists of androgen deprivation therapy (ADT) which aims to block production of testosterone and/or block its action on testosterone receptors. In prostatic tissue, testosterone acts on cells to promote growth and proliferation, blocking these signals with ADT leads androgen sensitive cells to undergo apoptosis³. When metastatic prostate cancer responds to ADT it is termed metastatic hormone sensitive prostate cancer (mHSPC). Whilst a good response to ADT is often seen initially, it is inevitable that the disease begins to progress despite treatment to become what is then termed metastatic castrate resistant prostate cancer (mCRPC)^3,4.

When VARIANT was conceived the standard of care for mHSPC, for most patients, was ADT alone, at the subsequent development of castrate resistance additional treatment would then consist of either next generation androgen receptor targeted agents (ARTAs), non-hormonal treatment with chemotherapy or drug delivered radiotherapy (radium 233)⁵. ARTAs such as abiraterone or enzalutamide are typically the preferred option as they generally have less side effects. However, response is variable with a proportion of patients resistant to the treatment primarily and all patients eventually becoming treatment refractory³. Predicting a positive response in individual patients is challenging and failed response, disease progression and uncertainty around treatment can be difficult for patients.

One method to better determine effective therapy for these patients has been proposed in the form of monitoring levels of androgen receptor splice variant 7 (AR-V7). Androgen receptor splice variants are variations of the androgen receptor protein which lack a portion of the normal ligand binding domain and allow signalling despite lack of activation by a binding ligand^6–9. AR-V7 is an example of one these variants and has been found to have a higher expression in prostate tissue of patients with mCRPC than in those who are hormone naïve and has a strong association with hormone resistance and metastatic disease^9,10. Moreover AR-V7, which can be detected on circulating tumour cells (CTCs) within patients’ blood samples, has been implicated in resistance to abiraterone and enzalutamide^11,12.

The VARIANT randomised controlled trial (RCT) was designed to assess the feasibility of utilising the AR-V7 biomarker in order to determine the optimal treatment pathway, treating with ARTAs in those patients likely to benefit and alternative treatment options to optimise disease control in patients in whom further hormonal treatments are likely to be futile. The primary objective was to establish feasibility in conducting a definitive randomised trial comparing AR-V7 biomarker-driven management with the current standard care in patients with mCRPC. The secondary objectives were to (1) estimate AR-V7 biomarker prevalence in the trial population to inform sample size calculations for a definitive randomised control trial; (2) assess recruitment, compliance and retention rates; (3) confirm outcome measures for a future definitive trial and establish trial data response rates, variability, and data quality; and (4) establish a blood biobank to include baseline, 12 and 24-week blood samples for future translational studies.

We aim to report these results and also discuss the reasons why the trial was not successfully completed to target with a view to share lessons learnt from our feasibility study.

Methods

We conducted a randomised, open, feasibility trial, with participants recruited from three secondary care National Health Service (NHS) organisations in the UK: Velindre University NHS Trust, The Newcastle upon Tyne Hospitals NHS Foundation Trust and NHS Greater Glasgow and Clyde. This was registered with the ISRCTN trial registry on 12/08/2019, with the identifier: ISRCTN10246848 available at https://doi.org/10.1186/ISRCTN10246848. Favourable ethical opinion was obtained from the Wales National Research Ethics Service (NRES) Committee, reference: 18/WA/0419.

Patients were identified from urology/oncology clinical services and were approached about the trial during their routine clinic appointments. To be eligible for the study, patients were aged ≥18 years old with mCRPC and high-risk features clinically suitable for ARTA or chemotherapy. The eligibility criteria is published in full in the trial protocol which is available as open access (https://pubmed.ncbi.nlm.nih.gov/31857319/)¹³. The criteria includes: disease progression despite medical or surgical castration, suitability for treatment with at least one ARTA and one non-hormonal therapy and at least two high risk features. High risk features were defined as: age <60 years at time of diagnosis of metastatic disease, bone metastases present at time of diagnosis, Gleason score 8–10, presence of visceral metastases, PSA doubling time of less than 3 months, elevated alkaline phosphatase, Eastern Cooperative Oncology Group (ECOG) Performance status worse than or equal to 1, previous treatment for CRPC with docetaxal chemotherapy or ARTA¹³.

The trial was designed as a feasibility trial according to the definition of Eldridge et al. (2016)¹⁴. Feasibility includes the deliverability of the intervention and in this case, assessment of the frequency of the positive assay measurements (predicted at approximately 30%). The target sample size was designed according to external pilot RCT recommendation by Teare et al. (2014)¹⁵ where it is recommended that data is collected on a minimum of 60 patients per arm to estimate an ‘event’ rate in a single treatment arm. We planned to calculate a pooled estimate of overall recruitment rate and overall biomarker prevalence rate with a planned recruitment target of 70 patients in total to allow for dropout.

The target was to recruit 70 patients from the three centres: Newcastle, Glasgow and Cardiff. Participants were randomised using a method of random permuted blocks of concealed variable block size and stratified by site in the ratio 1:1 to receive personalised standard treatment (intervention) or standard care (control). In the personalised standard treatment group, participants’ treatment was guided by the results of the AR-V7 biomarker test. Participants randomised to the control arm received standard care without biomarker guided treatment. Details of the protocol for the blood sampling, processing and analyses are previously published¹³. In brief, 2 x 10ml blood samples using ACD-A Blood Collection Tubes were collected at baseline, 12 and 24 weeks. These samples were sent to Newcastle University on the day of collection using a courier service and shipped at a temperature below 10°C, i.e. on cool packs, but not frozen. These samples were used for CTC and cfDNA analysis and to provide the AR-V7 biomarker result using a validated commercially available kit - AdnaTest ProstateCancerPanel AR-V7 circulating tumour cell (CTC) quantitative RT-PCR (RT-qPCR) assay (Qiagen®)¹⁶.

Outcome measures were defined as feasibility measures to inform the definitive RCT and clinical measures. Feasibility measures included recruitment rates, proportion of patients who were eligible, the proportion who agreed to be randomised, the baseline prevalence of AR-V7 expression, how assessable blood samples were for the biomarker and timeline involved in processing the samples and whether patients were compliant with the recommended treatment and completing study measures. Clinical measures included time to prostate specific antigen (PSA) progression, clinical progression, cancer specific and overall survival. Clinical progression could be determined as a result of progressive symptomology or radiologically. As VARIANT was a pragmatic trial the latter was as determined by local radiology or multidisciplinary team.

Further to this quality of life (QOL) was assessed at baseline, 12 weeks and 24 weeks using the validated EORTC quality of life cancer questionnaires (QLQ-C30) with additional prostate cancer specific module (QLQ-PR25) (https://qol.eortc.org/). Additionally, a short non-validated ‘Use of Health Services Questionnaire’, consisting of ten questions assessing how patients utilised health resources during the trial was completed once at the end of the trial, to aid in future health economic evaluation, this is available as extended data¹⁷.

Further information about the methods including detailed eligibility criteria and outcomes is available in the earlier published protocol¹³.

Results

Recruitment, eligibility, randomisation and baseline demographics

Participant flow is summarised in the Consolidated Standards of Reporting Trials (CONSORT) diagram in Figure 1, additionally the CONSORT checklist is available as extended data^17,18. Of the 14 patients who were assessed against eligibility criteria, two patients were excluded as they were deemed too unfit to participate. Of the remaining 12 patients, all 12 agreed to be randomised with six patients randomised into each arm, four of these patients were randomised to have validation blood samples sent to the Cardiff labs. No participants withdrew or were lost to follow up over the course of the study. Baseline demographics are provided in Table 1.

Figure 1. CONSORT flow diagram.

Table 1. Clinical demographics and medical history of participants at screening.

		Personalised Treatment (n = 6)	Standard Care (n = 6)
Disease characteristics (at initial diagnosis)
PSA	median (range) (ng/ml)	116.7 (14.1–436)	16.25 (1.7–991)
Gleason score	6 | 7 | 8 | 9 | 10	0 | 1 (17%) | 2 (33%) | 3 (50%) | 0	1 (17%) | 0 | 2 (33%) | 3 (50%) | 0
TNMa Stage	T1 | T2 | T3 | T4	0 | 0 | 5 (83%) | 1 (17%)	0 | 1(17%) | 0 | 5 (83%)
	N0 | N1	3 (50%) | 3 (50%)	4 (67%) | 2 (33%)
	M0 | M1	1 (17%) | 5 (83%)	4 (67%) | 1 (33%)
ECOGb PSc	0 | 1	5 (83%) | 1 (17%)	4 (67%) | 2 (33%)
At entry to Variant
Age	median (range) years	70.2 (62.0–76.9)	66.4 (55.4–74.2)
Time since diagnosis	median (range) years	1.4 (0.79–11.5)	3.1 (0.95–8.77)
Metastatic disease location	Bone Visceral Lymph Node	4 (67%) 2 (33%) 2 (33%)	4 (67%) 0 (0%) 2 (33%)

^aTumour, Nodes and Metastases, ^bEastern Cooperative Oncology Group ^cperformance status

Recruitment issues and trial end

Due to delayed site opening, the recruitment period lasted less than nine-months rather than the full 12-month recruitment period that was planned. The opening of the three sites was projected to complete by the start of October 2018, however the first site started recruiting in July 2019 and all three sites were not fully operational until February 2020. This timeline and associated recruitment rates are shown in Figure 2. Recruitment rates were significantly lower than expected with an average of just over one participant recruited per month over the nine-month recruitment period, in contrast to a target of six participants per month. This was due to multiple reasons including slower timelines for sites opening to recruitment than originally planned and changes in clinical management pathway. The clinical management of metastatic prostate cancer evolved during study set up and early recruitment. This included the up-front use of docetaxel, though off license at the time, recommended in new NICE guidance published in May 2019¹⁹. As a result the management of men with metastatic prostate cancer now involves treatment with either novel hormonal therapies and/or chemotherapy prior to the onset of castration resistance and the conventional mCRPC stage we were examining is now uncommonly seen.

Figure 2. Recruitment timeline by month and site, whereby green shading indicates open for recruitment.

In addition to the above, further delays were caused by regulatory approval, this was secondary to the decision by the European Medicines Agency to restrict the use of radium-223, one of the non-hormonal treatment options²⁰. The AR-V7 assay being used in the study also changed (non-CE marked). Both of these changes required review of regulatory requirements and a delay in submission of ethics approval, pushing back all subsequent milestones.

The trial management team explored alternative measures to increase the rates of recruitment including the addition of extra sites. Southampton Hospital and University Hospital Bristol had been approached to take part however, the changing clinical management pathways and other competing studies meant that the new sites would face similar, if not the same, issues faced at the initial three recruiting sites. Nevertheless, we were able to demonstrate that some men with advanced cancer were willing to randomise to a study of biomarker-directed therapy

Following a Trial Oversight Committee meeting, recruitment to the trial was halted on 18^th March 2020 at all sites. As a result of the COVID-19 pandemic (also March 2020), the central Newcastle University laboratories were forced to close. No blood samples were collected from participants for their follow up visits; however, participants were asked to complete the questionnaires remotely (sent to them via post or completed with a research nurse over the phone).

Discussion

VARIANT successfully recruited 12 men to be randomised for biomarker guided treatment for prostate cancer and followed them up for 24 months. Though it failed to recruit as planned, this doesn’t appear to be due to a lack of willingness by clinicians or patients, with only two screen failures reported within the study, both secondary to patient fitness. The aforementioned delays impacted negatively with our ability to keep up with the rapidly changing field of prostate cancer management; we were aware of the potential for changes in treatment practice but had expected to fully recruit before these were realised. In the last few years there have been major changes in clinical practice with results from multiple clinical trials. For example, STAMPEDE, GETUG and CHAARTED compared ADT in mHSPC to ADT combined with docetaxel and found the addition of docetaxel up front led to an overall survival advantage^21–23. In time this led to newly diagnosed mHSPC patients being treated with ADT + Docetaxel if fit enough. Multiple trials investigating the role of ADT and ARTA (with or without chemotherapy) have now also published their results leading to further direct changes in both the management of patients with mHSPC and an indirect shift in the care of these men when they develop mCRPC^21,24–27. These changed the management pathway of the patient cohort selected for this trial, as treatment at the time of development of castrate resistance is dependent on the prior treatment given²⁸. Though this issue is not intrinsic to VARIANT, it did influence our ability to recruit participants, and changed the validity of the underlying hypothesis of the trial. Another factor was the assay reproducibility failures that hindered our initial AR-V7 status reporting. This was quickly recognised due to our planned cross-site validation of blood samples and the cause identified. We would highlight the importance of cross site validation for future biomarker trials along with ensuring adequate time and funding is allocated to testing assay reliability prior to commencement.

Clinical trials not meeting their objectives is by no means uncommon, with results often going unpublished. One study reviewing trials of 640 novel therapeutics found that 344 did not continue in clinical development and of these only 40% had their results published in peer reviewed journals²⁹. Whilst the most common cause of difficulties experienced in those trials for novel therapeutics was inadequate efficacy, our experience with unsuccessful recruitment was found to be the most common identified within both urology and oncology trials^30–32. Bandari et al. identified 1340 clinical trials in urology over a 10-year period of which 618 were unsuccessful, 41% of there were attributed to poor accrual, other causes included inadequate budget (9%), sponsor cancellation (7%) and poor interim results (7%). Within urology trials a significant association was found between unsuccessful trials and trials within oncology or andrology, device trials and trials funded by a combination of government and industry grants³⁰. Furthermore, a study in the UK across all specialities looking at trials funded by the MRC and HTA between 1994–2002 found that only 31% of studies recruited 100% of their original target and 45% achieved <80% of their original target, with 30% of trials reducing their recruitment targets and 54% requesting a trial extension³³.

Within VARIANT recruitment was well below the estimated level, with one site not recruiting a single patient. Successful recruitment has previously found to be associated with trial sites with prior track record of successful trials and also trial staff enthusiasm³⁴. Levitt et al. looked at recruitment levels across a number of sites in a large perinatal trial and identified factors associated with improved recruitment. They found that clearly defined recruitment systems, staff engagement, having a dedicated and experienced trial coordinator and a shorter time taken from ethics approval to first recruit were all associated with above average recruitment³⁵. They concluded that it may be better to focus resources on fewer sites with adequate resources and engaged staff³⁵. A formal process evaluation, such as the Quintet (Qualitative Research Integrated within Trials) recruitment intervention (QRI), could have helped identify barriers to recruitment, but was not part of the funded protocol. For feasibility studies, where there are predicted concerns about recruitment, a QRI to explore barriers and develop plans to optimise recruitment could be useful³⁶

Another method to try and improve trial success is the use of adaptive trial design whereby outcomes are assessed at pre-defined points and can be modified based on pre-specified rules. As a result, use of resources can be more efficient and potentially fewer patients may be required^37,38. One such example of this in urology is the STAMPEDE trial, briefly mentioned earlier, which examines systematic therapy in advancing or metastatic prostate cancer^21,39. Another technique being assessed to improve trial design and increase success rates is artificial intelligence. Proposed applications include machine learning techniques used to enhance patient recruitment through automatic eligibility assessment and trial recommendation⁴⁰.

AR-V7 remains clinically relevant with a recent systematic reviews finding a positive AR-V7 status to be associated with a reduced overall survival (OS) in comparison to AR-V7 negative patients^41–43. This was the case for both ARTA treatment and chemotherapy, although to a lesser extent in the latter. Where treatment response was compared chemotherapy was associated with a superior survival in AR-V7 positive patients than those treated with ARTA, this difference was not observed in those who were AR-V7 negative^42,43. Whilst some studies have continued to examine its use as a biomarker and further develop assays other studies are exploring the means to directly target the AR-V7 variants to overcome hormone resistance^44–47.

Conclusions

We present the results of the VARIANT clinical trial looking at the AR-V7 biomarker to guide treatment for patients with mCRPC. We can conclude that some men with prostate cancer are willing to take part in trials utilising biomarker guided treatment. However, due to issues with recruitment secondary to unforeseen delays and change within the management of prostate cancer the trial did not complete as planned. The lessons learned from this pilot trial are applicable to other research particularly in relation to fields where there is a rapid advance in knowledge.

Abbreviations

CRPC: castrate resistant prostate cancer

AR-V7: androgen receptor splice variant 7

NHS: National Health Service

ADT: androgen deprivation therapy

mHSPC: metastatic hormone sensitive prostate cancer

mCRPC: metastatic castrate resistant prostate cancer

ARTA: androgen receptor targeted agents

CTCs: circulating tumour cells

RCT: randomised control trial

PSA: prostate specific antigen

QOL: quality of life

CONSORT: Consolidated Standards of Reporting Trials

ECOG: Eastern Cooperative Oncology Group

PS: Performance Status

TMG: trial management group

Consent

Written informed consent for publication of the patients’ details was obtained from the patients.